Nitric oxide function in atherosclerosis

Atherosclerosis is a chronic inflammatory process in the intima of conduit arteries, which disturbs the endothelium-dependent regulation of the vascular tone by the labile liposoluble radical nitric oxide (NO) formed by the constitutive endothelial nitric oxide synthase (eNOS). This defect predisposes to coronary vasospasm and cardiac ischaemia, with anginal pain as the typical clinical manifestation. It is now appreciated that endothelial dysfunction is an early event in atherogenesis and that it may also involve the microcirculation, in which atherosclerotic lesions do not develop. On the other hand, the inflammatory environment in atherosclerotic plaques may result in the expression of the inducible NO synthase (iNOS) isozyme. Whether the dysfunction in endothelial NO production is causal to, or the result of, atherosclerotic lesion formation is still highly debated. Most evidence supports the hypothesis that constitutive endothelial NO release protects against atherogenesis e.g. by preventing smooth muscle cell proliferation and leukocyte adhesion. Nitric oxide generated by the inducible isozyme may be beneficial by replacing the failing endothelial production but excessive release may damage the vascular wall cells, especially in combination with reactive oxygen intermediates.

Atherosclerotic lesions develop in the inner coat (or tunica intima) of the aorta, the large elastic arteries e.g. the carotid arteries and the arteries supplying the lower extremities, and the medium-sized muscular arteries, such as the coronary arteries. At birth, the intima consists solely of endothelial cells, but soon after birth focal and circumferential thickening occurs. 1 This spontaneously developing intima consists of smooth muscle cells, connective tissue and isolated macrophages, and is considered an adaptation to mechanical wall stress. 2 Although not pathologic at this stage, the thickened intima marks locations where atherosclerosis tends to develop later in life under the in uence of atherogenic stimuli e.g. hypercholesterolaemia.

Features of human atherosclerosis
Early atherosclerotic lesions are characterized by the deposition of lipids and the appearance of macrophages and T-lymphocytes in the intima. As macrophages and a few smooth muscle cells underneath the endothelial cells accumulate lipid, they acquire a 'foamy' appearance. Clusters of lipid-laden cells become macroscopically visible as fatty streaks. 3 Progressively, these at, fatty lesions transform to raised brolipid plaques, as intimal smooth muscle cells proliferate and deposit extracellular matrix, mainly collagen. In a subsequent stage, the advanced lesion has a characteristic microanatomy with a core of extracellular lipid separated from the media by smooth muscle cells and covered at the luminal side by a thick brous cap. Surrounding the lipid core are lipid-lled foam cells. The ischaemia in the necrotic core initiates angiogenesis. This type of plaque may cause narrowing of the lumen once the compensatory vascular remodelling process which increases the external diameter of the vessel becomes exhausted. Only then, lesions become angiographically visible. The nal stage, the complicated plaque, may arise either from ssure of the brous cap or from intra-plaque haemorrhage. The thromboembolic events following plaque ssure are a major cause of clinically manifest acute ischaemic syndromes. If the thrombus is not occlusive, it becomes incorporated into the plaque and is organized by invading macrophages and smooth muscle cells, thereby further compromising the lumen of the vessel. The sequence of ssure, thrombus formation, organization and incorporation into the plaque may occur repeatedly.

Models o f atherosclerosis
Current knowledge of the initiation of the atherogenic process is largely based on rabbit or primate models of hypercholesterolaemia, which may be diet-induced or genetically determined as in Watanabe heritable hyperlipidaemic (WHHL) rabbits. Hypercholesterolaemia provokes intravascular lipid in ltration leading to the formation of fatty streaks, which resemble early human lesions. 3 Protracted cholesterol feeding eventually results in advanced brolipid plaques containing necrotic debris, as in advanced human disease.

Models o f intim al thickening
Intimal thickening can be induced experimentally by creating a modest mechanical injury of the smooth muscle cells of the media. The most extensively investigated model involves balloon denudation of the intima of the rat carotid artery with an embolectomy catheter. 4 The discrete mechanical injury of the underlying media evokes smooth muscle cell proliferation in the media, followed by migration to the intima and an extended phase of intimal proliferation. The endothelial cells are completely removed by the initial insult and regrowth of the endothelial cells from the lesion edges is virtually absent. The removal of the endothelial cells is not essential nor suf cient for the process of intimal hyperplasia.
Placing a exible collar around the rabbit carotid artery does not create direct endothelial injury, but induces smooth muscle cell proliferation in the media, followed by migration and prolonged proliferation in the intima. 5 Both models illustrate the three wave paradigm for the involvement of smooth muscle cells in the formation of intimal cushions. 4 The in ation of an angioplasty balloon in arteries of rabbits, pigs or other experimental animals is used to mimic restenosis due to accelerated intimal thickening after percutaneous transluminal coronary angioplasty (PTCA). The vessel wall distension by the repeated in ation of a slightly oversized balloon creates a much more extensive injury of the media and the lamina elastica interna than the gentle passage of an embolectomy catheter. Unlike balloon denudation, the balloon angioplasty thus predisposes to thrombus formation. In accordance with restenosis after PTCA in humans, the incorporation and organization of the non-occlusive thrombus adds to the bulk of neointima formation. 4 A further difference with balloon denudation is the quick and often complete recovery of the endothelial cell layer through outgrowth from patches of cells which remained present after the angioplasty.

Atherosclerosis is a Chronic In¯ammatory Process
The long-standing and continuously re ned 'response-to-injury' hypothesis 6 considers the lesions as the result of an excessive in ammatory-broproliferative response to various forms of insults to the endothelium and smooth muscle. Moreover, the presence of T-lymphocytes in atherosclerotic lesions at all stages of development points to an important immunologic component in atherogenesis. 7 T-lymphocytes and macrophages are capable of producing numerous in ammatory mediators and growth factors, and have been demonstrated to be in an activated state in atherosclerotic lesions.

Pathogenetic mechanisms in hypercholesterolaemia-induced atherogenesis
Several different sources of injury to the endothelium can lead to endothelial dysfunction and initiate the disease process. 6 In hypercholesterolaemia-induced atherosclerosis, the major causal agent is now assumed to be oxidized LDL (oxLDL). 8,9 Oxidation of lipoproteins ooding the intima may result from the production of reactive oxygen intermediates or 15-lipoxygenase activity in the endothelial cells (Fig. 1). OxLDL in turn is cytotoxic to endothelial cells by the metal-catalysed production of free radicals from lipid hydroperoxides contained in the modi ed lipoprotein particle. 10 Furthermore, oxLDL is chemotactic for monocytes and Tlymphocytes. Newly formed epitopes in oxLDL elicit cell-mediated and humoral immune re-sponses. 7 Minimally oxidized LDL stimulates the endothelial cells and smooth muscle cells to secrete monocyte chemotactic protein-1 (MCP-1) and growth factors involved in the differentiation and proliferation of monocytes, and oxLDL may, synergistically with cytokines, promote mononuclear leukocyte adhesion to the endothelium through the induction of vascular cell adhesion molecule-1 (VCAM-1). 11,12 The oxidative stress-sensitive nuclear transcription factor B (NF-B) may be a crucial intermediate in the in ammatory activation of the endothelium. 13 Monocyte-derived macrophages internalize oxLDL through scavenger receptors. As these receptors are not down-regulated by the intracellular cholesterol level, massive cholesterol accumulation occurs and the macrophages transform to foam cells.
Thus, it appears that endothelial cells, through the oxidation of LDL, recruit macrophages to remove the invaded lipoprotein particles. This attractive hypothesis also implies that a chronic in ammatory response will develop if the macrophages are unable to eliminate oxLDL suf ciently.

Atherosclerosis and Nitric Oxide Signalling
The nitric oxide signalling pathway in normal arteries A major step forward in the understanding of blood vessel physiology was the discovery by Furchgott and Zawadzki 14 of a factor released by the endothelium that relaxed the underlying smooth muscle. This endothelium-derived relaxing factor (EDRF) was later identi ed as nitric oxide (NO) 15 17 or a related nitrosylated compound e.g. S-nitrosocysteine. 18,19 Nitric oxide is formed by a ve-electron oxidation of a terminal guanidino nitrogen atom of the amino acid Larginine, with concomitant formation of L-citrulline, by an enzyme known as nitric oxide synthase (NOS). 20 There are two major classes of NO synthases: constitutive and inducible (iNOS) enzymes. Constitutive isoforms are expressed in endothelial cells (eNOS), in neuronal cells (nNOS) and in certain other cell types. The activity of these isoforms is strictly calciumcalmodulin dependent and is present both in the cytosol and associated with membranes. OxLDL promotes mononuclear cell adhesion, in® ltration and proliferation by stimulating the endothelial cells to express adhesion molecules, e.g. VCAM-1, and to produce chemotactic factors, e.g. MCP-1, and growth factors. Uptake of oxLDL by macrophages leads to foam cell formation.

Mediators of In¯ammation´Vol 6´1997 5
Stimulation of the appropriate receptors on the endothelial cell by physical (shear stress resulting from increased ow, mechanical deformation) or chemical (acetylcholine, bradykinin, substance P, ATP) stimuli raises the cytoplasmic calcium levels, with concomitant eNOS activation and formation of NO. 21 Serotonin (5-hydroxytryptamine, 5-HT), although a potent vasoconstrictor through activation of 5-HT 2 receptors on the vascular smooth muscle cells, also mediates dilatation by the EDRF/NO-dependent mechanism, through activation of 5-HT 1like receptors on the endothelium. 22 In biological systems, the dominant reactions of NO will be with another free radical such as superoxide anion, transition metals such as haem iron, or oxygen. 20 In the vessel wall, NO diffuses into the underlying smooth muscle cells to react with the haem group of a cytoplasmic guanylate cyclase. The formation of cyclic GMP then causes vasodilatation. 23 25 Nitric oxide also raises cyclic GMP in the endothelial cells themselves, which inhibits the production of the potent endotheliumderived contracting factor endothelin. 26,27 Inhibitors of NOS and guanylyl cyclase revealed an inhibitory role of NO but not of cyclic GMP in endothelin secretion by porcine aortic endothelial cells. 28

Nitric oxide signalling in atherosclerosis
It has been recognized for a long time that atherosclerotic blood vessels are very susceptible to the development of vasospasm in vivo 29 32 and are hyperreactive to contractile agonists in vitro. 32 35 Because coronary vasospasm can be provoked by several stimuli with different mechanisms of action, it has been proposed that dysfunction or denudation of the endothelium in atherosclerosis may contribute to that phenomenon by leaving constrictor responses unopposed. 36 38 Even before it was realized that NO accounts for the biological activity of endothelium-derived relaxing factor, it was indeed demonstrated that artery segments obtained from atherosclerotic animals showed a loss of endothelium-dependent relaxation in organ bath experiments. 35 While the expected hypertensive effect might contribute to the progression of cardiovascular disease, this also implies that, besides a dysfunctional endothelial NO pathway, other factors are involved in the initiation and progression of atherosclerotic plaques, since lesions do not develop in these microvessels. The systemic nature of the endothelial dysfunction could be of use in the non-invasive evaluation of endothelial function in readily accessible arteries. 96 In summary, established atherosclerosis or the presence of risk factors e.g. hypertension, hypercholesterolaemia, and even male gender, 97,98 decrease the activity of the EDRF/NO pathway. Endothelium-dependent dilatation is lost progressively as atherogenesis continues. Conduit vessels with lesions as well as resistance vessels in which lesions do not develop are affected, the latter apparently to a lesser extent. 65,93 Explanations for the Defective NO Signalling Pathway The mechanisms underlying the dysfunctional endothelial NO signalling pathway in atherosclerosis and hypercholesterolaemia are multifactorial (reviewed in Refs 99 and 100). Atherosclerotic arteries demonstrating disturbed endothelium-dependent relaxation are still capable of dilating to the NO donor nitroglycerin which provides the smooth muscle with NO upon enzymatic or thiol-dependent bioconversion. 101 As this demonstrates that the smooth muscle is still responsive to the dilatory action of NO, defective endothelial EDRF/NO release or increased NO inactivation after release appear to be involved. This is supported by the decreased release of bioactive NO from isolated perfused atherosclerotic arteries as assessed by a superfusion bioassay. 41,44,50,102,103

Endothelial receptor dysfunction
Endothelium-dependent dilatation is lost in a progressive, hierarchical fashion. 82 Vasodilator responses to acetylcholine and serotonin are lost early, before impairment of the dilatation to other receptor agonists e.g. substance P, to the receptor-independent stimulus calcium ionophore A-23178 or to mechanical stimuli. The agonist speci city of the early dysfunction suggests that it is not caused by a nonspeci c impairment of the ability of the endothelial cells to produce NO, but points to selective alterations in endothelial receptor function or postreceptor effector pathways. This view is strengthened by the observation that the receptor-induced release of other endothelial products, such as prostacyclin, is attenuated as well. 104,105 In this respect, it has been demonstrated that the pertussis toxin-sensitive G i protein signalling pathway, which is employed by serotonin to elicit endothelium-dependent relaxation in the pig coronary artery, is impaired in the early stages of the atherosclerotic process. 106 Several studies have demonstrated that incubation of vessel segments with lipoproteins, and in particular with oxLDL, inhibited endothelium-dependent relaxation in a way similar to hypercholesterolaemia in vivo (reviewed by Flavahan 99 ). Lysophosphatidylcho-line, a component of oxLDL, mimicked the effects of the whole particle, 107 109 inhibited agonist-stimulated calcium signalling in cultured endothelial cells 110 and selectively inhibited G i protein-dependent signalling in porcine endothelial cells. 111

Expression of eNOS activity
The receptor selectivity (see above) argues against a reduced expression of eNOS activity in endothelial cells overlying atherosclerotic lesions. This assumption has recently been con rmed by in situ hybridization of eNOS mRNA and immunohistochemistry of eNOS protein in the aorta of hypercholesterolaemic rabbits. The results suggested that the expression of eNOS mRNA and protein was even increased in endothelial cells overlying brofatty plaques. 112 Studies of the expression and the activity of eNOS after in vitro exposure of endothelial cells to LDL, oxLDL or cholesterol yielded contradictory results, ranging from initial upregulation, via no effect to downregulation of eNOS expression. After acute exposure of the isolated rabbit carotid artery to cholesterol-rich liposomes, acetylcholine-induced release of EDRF/NO was evaluated functionally in a superfusion bioassay and appeared to be enhanced. 113 This could be due to augmentation of the release and/or prolongation of the halflife of NO. Exposure of endothelial cell membranes to liposomal cholesterol raised the activity of plasma membrane bound eNOS at low cholesterol concentrations, but had the opposite effect at higher concentrations. 114 The effects were attributed to modulations of the lipid environment of the membrane bound eNOS. Cholesterol was without effect on the activity of eNOS in the cytosol of endothelial cells. Interestingly, the increased activity of particulate eNOS was accompanied by a concentration-dependent increase in superoxide anion production, but the authors did not investigate whether eNOS was the source (see below). Low concentrations of oxLDL have been reported both to increase 115 and to decrease 116 the expression of mRNA, protein and activity of eNOS in cultured endothelial cells. The upregulation of eNOS mRNA was mimicked by lysophosphatidylcholine, one of the many constituents of oxLDL. This discrepancy between both reports could be due to large variability among different preparations of oxLDL with respect to biological activities. Downregulation of the expression of mRNA of eNOS by higher concentrations of oxLDL appears to be a more consistent nding. 115,116 Native LDL was without effect on eNOS mRNA levels and eNOS activity, 115,116 although exposure of cultured endothelial cells to LDL may promote superoxide anion formation by eNOS 117 (see below).

Arginine availability
Although arginine availability seems to be sufcient initially in view of the receptor-selectivity of the endothelial dysfunction (see above), studies on endothelium-dependent vasodilatation suggest that L-arginine depletion may occur. It should be noted, however, that the studies addressing the effect of the NO precursor show discordant results. In this respect, the behaviour of conduit arteries with overt atherosclerosis appears to be different from arterioles in the microcirculation, in which atherosclerosis does not develop.

Co nduit arteries with atherosclerosis
Most authors agree that in vitro L-arginine addition fails to restore the endothelium-dependent relaxations in the aorta 118 120 or femoral artery 121 with cholesterol-induced atherosclerotic lesions. One report showed that acute in vivo L-arginine administration to hypercholesterolaemic rabbits improved the endotheliumdependent relaxations in isolated large vessels in vitro, but it should be noted that responses to nitroglycerin were affected to a very similar extent. 122 Also prolonged in vivo L-arginine treatment ameliorated endothelium-dependent relaxations of isolated segments only marginally. 123 As the endothelial dysfunction is strictly dependent on the size of the lesions in rabbit conduit arteries, 35,49,124,125 the marked antiatherogenic effect of prolonged L-arginine supplementation 123 (see below) most likely explains the improved endothelium-dependent relaxations. In patients with coronary or peripheral artery occlusive disease, a positive effect of Larginine on endothelium-dependent dilatation of the conduit arteries was lacking. 126,127 Co nduit arteries witho ut overt atherosclerosis Although the rabbit basilar artery develops neither atherosclerotic lesions nor a clear endothelial dysfunction after prolonged hypercholesterolaemia, 49 an improvement of the endothelium-dependent relaxations has been reported after in vitro exposure to L-arginine. 128 In addition, L-arginine also attenuated the augmented vasoconstrictor responses to potassium chloride, serotonin and endothelin. The authors suggested that the normalization by L-arginine of both the endothelium-dependent relaxation and the constrictor responses was the result of increased EDRF production. However, as cyclic GMP-mediated relaxation induced by endothelium-independent agonists was not studied, and as the contraction to a depolarizing potassium chloride solution is not affected by basal EDRF release, 37 it is not entirely clear whether the actions of L-arginine can be attributed solely to enhanced endothelial NO production. Lefer and Ma measured constrictions evoked by the NOS inhibitor L-NAME as an index of basal NO release by the endothelial cells of rabbit coronary arteries isolated after three weeks of cholesterol diet. 60 A reciprocal relationship existed between L-NAME evoked contractions and plasma cholesterol, suggesting that basal NO release by the segments became compromised in the absence of overt atherosclerosis. In vitro addition of L-arginine almost totally restored this index of basal NO production. However, as nonendothelial iNOS may be induced in arteries of cholesterol-fed rabbits, 129 it cannot be excluded that the vasoconstrictor responses to L-NAME resulted from inhibition of iNOS rather than eNOS.

Arterioles without overt atherosclerosis
The results obtained in conduit arteries suggest that L-arginine may upregulate the impaired eNOS activity only if the cholesterol-exposed arteries are still lesion-free. Accordingly, all studies, except one, 130 reported that L-arginine infusion resulted in marked improvement to complete restoration of endothelium-dependent vasodilatation in the coronary and peripheral microcirculation in hypercholesterolaemic rabbits, 131 pigs 88 and humans. 126,132 The mechanism of the amelioration of endothelium-dependent relaxations by L-arginine is not yet clear, and could be due to an interaction with smooth muscle cells or other effects. In view of plasma levels of L-arginine in the range of 150 to 250 mM and a K m of 5 to 10 mM for NOS isoforms, it is indeed surprising that Larginine availability can ever limit NO biosynthesis. 20 L-arginine enters cells by facilitated diffusion via the y transporter. 133 As exogenous L-arginine addition neither induces endothelium-dependent relaxations by itself, nor enhances agonist-induced endothelium-dependent relaxations in normal isolated vessel rings, 120,129,134 the intracellular stores appear to be suf cient for maximal eNOS activity in physiological circumstances. The increase in membrane cholesterol associated with hypercholesterolaemia might impair endothelial Larginine transport, thus eventually depleting the intracellular stores. The latter may also result from the increased output of inactive nitrogen oxides, as demonstrated in hypercholesterolaemic rabbit aorta. 103 However, reversal by Larginine of hypercholesterolaemic endothelial dysfunction may not simply re ect the replenishment of the substrate for NO production. The observation that the effect of L-arginine administration to hypercholesterolaemic rabbits is not sustained and depends on the anatomic site and sex 135 indeed supports a more complex mechanism of action. As the best results are obtained in the microcirculation after L-arginine treatment in vivo, other less well characterized systemic effects of the amino acid e.g. its secretagogue effects on the adrenals and pituitary gland, may prevail. 133 This is illustrated by observations in healthy persons, where L-arginine infusion stimulated basal and acetylcholine-induced relaxation in the peripheral circulation 130,136,137 and decreased the systemic blood pressure. 127,138 The concomitant increase in urinary nitrate and cyclic GMP could not simply be attributed to a direct stimulating effect of L-arginine on eNOS, as prostaglandin E 1 -induced dilatation also increased these parameters in the urine. 127 Furthermore, intravenous L-arginine administration increased urinary ow, which by itself resulted in enhanced excretion of nitrate and CGMP, in the absence of elevated nitrate plasma levels. 138 Endothelial NO synthase inhibition L-arginine may be effective in conditions where endogenous NOS inhibitors are formed. N G ,N G dimethylarginine (DMA) has been found in the urine and plasma of humans and inhibited macrophage and vascular NO synthesis in vitro and in vivo in animals and humans, 139 suggesting the existence of endogenous mechanisms to regulate NO synthesis. Recently, DMA was reported to be increased in the serum of cholesterol-fed rabbits. 140,141 All classes of NO synthases are liable to feedback inhibition by NO, 142 probably by the interaction of NO with the haem prosthetic group. 143 Hence, high output NO production by iNOS (see below) might downregulate eNOS activity. This is supported by the observation that chronic in vivo administration of large doses of an NO donor to rabbits depressed the ex vivo output of EDRF/ NO in response to acetylcholine, as assessed by means of bioassay. 135 Endothelial NOS may also be suppressed by other locally produced in ammatory mediators e.g. the T-lymphocyte-derived mediator interferon-c. 144

Inactivation of NO by superoxide anion
Superoxide anion is known to inactivate EDRF/ NO. 145,146 Generation of superoxide anion in situ in normal vessels reduced endotheliumdependent relaxation. 147,148 Under normal conditions, inactivation of EDRF by superoxide radicals is prevented by cytosolic CuZn superoxide dismutase (SOD) 149 and by extracellular SOD type C associated with heparan sulphate proteoglycans on the endothelial cell surface and in the interstitium. 148 Hypercholesterolaemia in the rabbit increased the intimal 103,150 production of reactive oxygen species, resulting in increased degradation of NO (reviewed by Harrison and Ohara 100 ). The tunica media beneath the atheromatous plaque in WHHL rabbits also inactivated EDRF/NO by an SOD-sensitive mechanism. 44 Increased vascular production of reactive oxygen species may result from enhanced xanthine oxidase activity in the endothelium 150 or from production by in ltrated monocytes. 151 In addition to direct inactivation of EDRF/NO by oxLDL and lysophosphatidylcholine, 152 oxLDL has been shown to stimulate the respiratory burst in neutrophils, 153 and lysophosphatidylcholine induced superoxide production in vascular smooth muscle cells via protein kinase C activation. 154 Endothelial NADPH oxidase systems, 155 activated by protein kinase C, 156 may also be involved. Protracted endothelial cell exposure to atherogenic native LDL concentrations increased superoxide anion production by three independent oxidative systems-cyclooxygenase, P450 isozyme and eNOS-of which the latter appeared to be the greatest source. 117 Nitrotyrosines, hallmarks of peroxynitrite formation from superoxide and NO, were detected intracellularly.
Furthermore, a striking feature of NOS is its ability to generate superoxide anion when either L-arginine or the cofactor tetrahydrobiopterin is limiting. 20 Under these circumstances, NADPH oxidation is uncoupled from synthesis of NO, and oxygen becomes the electron acceptor, resulting in superoxide formation. This has been demonstrated to occur in the constitutive NOS of the brain. Whether the low arginine levels needed for superoxide biosynthesis occur in intact endothelial cells in vivo is unclear. Arginine depletion of eNOS might occur from high local L-arginine consumption by iNOS (see below). Arginine availability may also be reduced by impediment of cellular uptake or delivery to eNOS, as has been suggested to occur in endothelial cell cultures treated with native LDL. 117 In the latter experiments, LDL-exposed cells produced signi cantly more superoxide anion than untreated cells, which was reversed by arginine supplementation. Inducible NOS did not seem to be involved, as Ca 2 -independent arginine-to-citrulline conversion under apparent V max conditions was low. Nevertheless, in conditions where iNOS, which is much more demanding for substrate then eNOS, is induced, insuf cient Larginine might result in superoxide anion release. This would also explain the bene t of providing L-arginine, i.e. to promote re-coupling, thus reducing vascular superoxide production and prolonging the half-life of EDRF/NO. These ndings provide new insight into the mechanisms by which hypercholesterolaemia might both stimulate superoxide production and decrease functional NO levels.
The disturbed balance between vascular superoxide and endothelial nitric oxide production, resulting in the loss of functional NO, may be compensated for by iNOS activity in the vascular wall (see below) and/or by upregulation of endogenous SOD. 157 159 Addition in the organ bath of CuZn SOD, which does not penetrate cells, or preincubation with extracellular SOD type C, which binds extracellularly to vascular structures, also protected against the detrimental effects of superoxide radicals on endothelium-dependent relaxation. 148 Conversely, exhaustion of these protective mechanisms, which may be time-, species-, or vesseldependent, may tip over the balance towards a net decrease in functional EDRF/NO.
In rabbits, but not in pigs, hypercholesterolaemia alone did not impair the endothelial dilator function in large vessels, but only occurred in arteries with intimal plaques, 35,44,46,48,49,125 with the exception of the coronary arteries. 60, 160 Apparently, the rabbit is capable of keeping the superoxide and nitric oxide production in balance, as long as lesions do not develop. Superoxide production in the media beneath the plaque 44 or the presence of fatty streaks containing large amounts of macrophages and lipids, may disturb the balance by the high local superoxide production and the trapping of the lipophilic NO molecule.
Raising the antioxidant capacity in the vessel wall by the administration of CuZn superoxide dismutase, polyethylene-glycolated 161 or liposome-entrapped 121 to ensure cell entrance, partly restored the endothelium-dependent relaxation in the isolated aorta of the cholesterol-fed rabbit. 121,161 In keeping with these ndings, it has been shown that addition of antioxidant vitamins in the diet of cholesterol-fed rabbits preserved the endothelium-dependent dilatation in the absence of an effect on lesion formation. 162,163 Also, dietary correction of hypercholesterolaemia in the rabbit normalized both the endothelial superoxide production and dramatically improved the vasodilator response to acetylcholine. 164 Oral administration of 2 g ascorbic acid produced marked improvement in the forearm vascular response to hyperaemia in patients with coronary artery disease. 165 However, short-term treatment with antioxidants of patients with hypercholesterolaemia did not improve the forearm vascular responses to acetylcholine. 166,167 Eventually, atherosclerotic plaques, in particular when lipid-rich, may trap NO and may also mechanically disturb the normal dilatation of the medial smooth muscle. At this stage, the relaxation to exogenous NO donors e.g. nitroglycerin, and to endothelium-independent dilator substances e.g. atrial natriuretic peptide also becomes impaired. 102

Atherosclerosis and Inducible Nitric Oxide Synthase Expression
Animal and human macrophages, 168,169 smooth muscle cells 170 172 and endothelial cells 173,174 are capable of expressing iNOS after stimulation with endotoxin or cytokines. In contrast to eNOS, iNOS produces high amounts of NO for a sustained period. 20 In early reports, the presence of a constitutive NOS in vascular smooth muscle cells has been suggested, 175,176 but these observations were probably related to the induction of iNOS during the isolation and manipulation of the cells or tissue. 172,177,178 Atherosclerosis and iNOS Only recently, functional and biochemical evidence suggested that cholesterol feeding of rabbits induced iNOS expression in the aorta 129 and in the lungs. 179 The addition of NOS inhibiting L-arginine analogues caused endothelium-independent contractions in the isolated atherosclerotic rabbit aorta, pointing to the continuous formation of NO by subendothelial iNOS. 129 The observation that the NOS inhibitors nitro-L-arginine methyl ester (L-NAME) and monomethyl-L-arginine (L-NMMA) were equipotent in this respect further supported the involvement of iNOS. 129 The expression of iNOS may account for the increased output of nitrogen oxides in arteries of cholesterol-fed rabbits. 103 Histochemical studies in WHHL rabbits con rmed the expression of iNOS in medial and intimal smooth muscle cells, and showed signi cant enhancement of endotoxin-induced iNOS expression in atherosclerotic rabbits compared with normal New Zealand White rabbits. 180 In a chronic rejection model of transplant atherosclerosis in the rat, both macrophages and smooth muscle cells stained positive for iNOS. 181 More recently, it has been reported that iNOS is present within human atherosclerotic lesions and co-localizes with nitrotyrosine in macrophages and smooth muscle cells. 182 Induction of iNOS was also observed in the endothelium and smooth muscle of intramyocardial vessels of patients with ischaemic heart disease. 183 The observation that iNOS induction in vascular smooth muscle cells, as in macrophages, is accompanied by upregulation of L-arginine transport, 184 may contribute to the stimulating effect of L-arginine on vessel relaxation in some experimental settings.

Mechanical injury and iNOS
The hypocontractility to several agonists observed after balloon denudation of rat 185,186 or balloon angioplasty of rabbit arteries 187 was also attributed to the induction of iNOS in the vessel wall and was already noticeable 6 h postinjury. 186 Unlike the case in normal arteries, Larginine evoked signi cant relaxation in deendothelialized balloon-injured vessel segments, which was reversed by the NOS inhibitor L-NAME. 187 In the balloon-injured rat carotid artery, reverse transcription and polymerase chain reaction ampli cation showed the appearance of iNOS mRNA already 24 h post surgery, and in situ hybridization located iNOS mRNA in neointimal smooth muscle cells, particularly at the luminal side of the vessel, conferring a nonthrombogenic surface. 188 Cytokines introduced in the affected vessel by in ltrating monocytes and T-lymphocytes may provide the stimulus for iNOS induction in the smooth muscle cells. Also, oxLDL 189 and LDL 190,191 have been shown to upregulate iNOS activity in macrophages and vascular smooth muscle cells under certain conditions. On the other hand, mediators that inhibit iNOS induction e.g. heat shock proteins 192 or NO itself, 193 may determine the nal output of NO.

NO: A Radical with Antiatherogenic Properties
Since the impairment in the EDRF/NO pathway occurs early or even precedes the development of visible lesions in the process of atherosclerosis, many authors have speculated on a causal role of this functional defect. This view is supported by a number of in vitro studies demonstrating the suppression by NO, produced endogenously or derived from NO donors, of several key processes involved in atherogenesis (Table 1).

Interference with oxid ative pro cesses
Since superoxide anion contributes to oxidative stress, LDL modi cation 194 and in ammatory gene transcription via the activation of NF-B, 13 the decreased formation or inactivation of superoxide by NO may be considered protective. In this respect, it has been shown that the NO derived from iNOS inhibits xanthine oxidase in interferon-c-stimulated macrophages 195 and that authentic exogenous NO inhibits xanthine oxidase in a cell-free system, 196 possibly by reversible alteration of the avin prosthetic site. 197 Nitric oxide also inhibits neutrophil superoxide anion production via a direct action on the NADPH oxidase. 196 The NO donors known as NONOates abrogate the cytotoxic effects of superoxide on Chinese hamster lung broblasts. 198 Moreover, NO also protected against cellular damage by other reactive oxygen species e.g. hydrogen peroxide and alkyl peroxides, by several mechanisms such as prevention of haem oxidation, inhibition of Fentontype oxidation of DNA, and abatement of lipid peroxidation. 199 Activation of the stress protein haem oxygenase by NO may contribute to its cytoprotective effect. 200 Furthermore, NO reduced the oxidative modi cation of LDL by macrophages, 201 204 endothelial cells 205 and lipoxygenase 206 by acting as a potent terminator of radical chain propagation reactions. 206 Also, as 15-lipoxygenase has been implicated in LDL oxidation, 8 NO might protect LDL by inhibiting lipoxygenase activity. 207 Conversely, the decreased expression of iNOS activity in oxLDLladen foam cells 208,209 has been implicated in the accelerated oxidation of LDL by these macrophages. 210 Furthermore, NO released by donor compounds inhibited the cellular toxicity of the lipid hydroperoxides contained in oxLDL, presumably by scavenging the propagatory free radicals generated during peroxidation of the endothelial cell membranes. 211 NO donors also blocked the hydrogen peroxide-related increase in endothelial permeability by a cyclic GMPmediated mechanism. 212

Interference with leuko cyte recruitment
Inhibition of NO synthase in endothelial cells by L-NAME increased the intracellular oxidative stress, resulting in enhanced adhesion of neutrophils via CD18/ICAM-1 interaction 213 or the upregulation of P-selectin on the endothelium. 214 Neutrophil adhesion to the endothelium was augmented by hypercholesterolaemia 60, 215 and this increase was prevented by the NO donor SPM 5185. 216 Inhibition of NO biosynthesis also induced the expression of VCAM-1 217 and upregulated MCP-1 mRNA and protein in cultured human endothelial cells, whereas addition of the NO donor SIN-1 dose-dependently decreased MCP-1 mRNA expression and secretion, presumably by suppressing a NF-B-like transcriptional regulator. 218 Authentic NO gas inhibited monocyte adhesion and chemotaxis 219 and exposure to shear stress inhibited monocyte adhesion by an NO-dependent mechanism. 220 Furthermore, NO donors decreased the cytokine-induced expression of the endothelial adhesion molecules VCAM-1, ICAM-1 and Eselectin. 217 Antiprolifer ative action of nitric oxide As the proliferation of vascular smooth muscle cells, macrophages and T-lymphocytes contributes to the progression of intimal lesions, cell growth inhibition by NO could signi cantly reduce lesion formation. Nitric oxide has been shown by several investigators to inhibit smooth muscle cell proliferation 221 227 as well as migration 228,229 in vitro. Both effects were cyclic GMP-mediated. T-cell proliferation is also reduced by NO. 230 232 Interestingly, the effect of NO appeared to be quite different in endothelial cells, in that it induced endothelial cell growth and motility in vitro and mediated the mitogenic effect of vascular endothelial growth factor. 233

Antipl atelet effects o f nitric oxide
Although the inhibitory effect of NO on platelet adhesion and aggregation 234,235 (reviewed by Bassenge 236 ) is often considered an anti-atherogenic effect of NO, platelets are minimally involved during the early stages of the atherogenic process. However, the endothelial surface over advanced human plaques often shows focal loss of cells 15 and platelet adhesion may promote the progression of those lesions, eventually leading to plaque ssuring and thrombosis. Also following gross mechanical injury evoked by balloon angioplasty, platelet-derived products have been proposed to contribute to neointima formation. 4

Inhibition of atherosclerosis by NO in vivo
Several in vivo studies (Table 2) support the concept that NO may suppress both atherosclerosis and intimal thickening. Oral L-arginine supplementation caused a striking inhibition of fatty streak formation in hypercholesterolaemic rabbits. 237 Since the adhesion of monocytes to endothelial cells is imperative to lesion formation in this model, the authors further addressed the effects of the NO precursor on the adhesion of a murine monocytic cell line to the aortic endothelium of cholesterol-fed rabbits ex vivo. 238 The enhanced endothelial adhesiveness for monocytes in hypercholesterolaemic aortas was signi cantly reduced if the rabbits had received supplemental dietary arginine and signi cantly increased in rabbits treated with L-NAME. This was associated with, respectively, increased or decreased elaboration of vascular nitrogen oxides as measured by chemiluminescence. The observation that hypercholesterolaemia-induced impairment in endotheliumdependent relaxation was only marginally improved by L-arginine treatment while lesion formation was signi cantly reduced, 123 suggests that the bene cial effect of L-arginine may primarily infer from the increased activity of iNOS. However, it should be noted again that arginine and other basic amino acids are potent hormonal secretagogues in adrenals and many other endocrine organs. 133 Hence, part of the anti-atherosclerotic effect of systemic arginine administration might be related to the release of glucocorticoids or other immunosuppressive hormones which are known to suppress intimal thickening and experimental atherosclerosis. The ambiguity of results obtained with Larginine is avoided in studies with NO donor compounds or NOS inhibitors. L-NMMA and L-NAME increased leukocyte adhesion in vivo by a CD11/CD18-dependent mechanism. 239 Conversely, the NO donor SIN-1 prevented leukocyte adhesion. The observation that both SOD and SIN-1 inhibited leukocyte adhesion only under conditions associated with superoxide formation suggests that the anti-adhesive properties of NO may relate to its ability to inactivate the superoxide anion. 240 Another NO donor attenuated leukocyte endothelial interaction in an in vivo model of ischaemia-reperfusion, and this appeared to be in part mediated through a decreased expression of endothelial P-selectin. 241 Pentaerythrityl tetranitrate, an organic nitrate, has been documented to inhibit cholesterol-induced fatty streak formation in rabbits, but the bene cial effect was not seen with isosorbide mononitrate, another organic nitrate. 242 This could be due to differences with respect to the development of tolerance or the NO releasing capacity between the two organic nitrates. Conversely, treatment with molsidomine, whose active metabolite is the sponta-neous NO donor SIN-1, actually enhanced lesion formation in the hypercholesterolaemic rabbit. 125 This may relate to the generation of superoxide anion from SIN-1, which could abrogate the bene cial effects of the simultaneously released NO.
Moreover, oral or parenteral treatment with the NO synthase inhibitor L-NAME for 4 to 12 weeks 243,244 enhanced fatty streak formation signi cantly. Therefore, the data suggest that vascular NO, produced by eNOS or iNOS, inhibits de novo formation of intimal lesions. However, this conclusion is somewhat confounded by the observation that L-NAME augmented plasma cholesterol levels in these rabbits, particularly after prolonged treatment. Since hypercholesterolaemia is the ultimate driving force for the lesions in this model, it is conceivable that this contributed to the accelerated atherosclerosis.
The inhibitory effect of NO on atherosclerosis may result from the above described in vitro actions, but NO-mediated decrease of endothelin production 28,245 may also be involved. Endothelin is a potent mitogen 246 and inducer of collagen synthesis 247 in vascular smooth muscle cell cultures, and its production may be increased in atherosclerosis. 248

Inhibition of intimal thickening by NO
The interferences with cholesterol absorption or metabolism are circumvented in studies of intimal thickening in animals with normal plasma cholesterol levels. Oral L-arginine supplementation suppressed intimal hyperplasia in experimental vein grafts 249 and after balloon denudation of the rat carotid artery. 250,251 The NOS inhibitor L-NAME reversed the effect of arginine, 250 indicating that the attenuation of the intimal hyperplasia was mediated by NO. The NO is presumably formed by iNOS, 185,186,188 since regrowth of the endothelial cells is virtually absent, and eNOS activity does not recover. Moreover, the endogenous biosynthesis of NO appears to modulate the process, since systemic 250 or local, perivascular 251 administration of L-NAME aggravated intimal thickening in response to balloon denudation. Increasing the ow in the injured carotid artery by ligating the contralateral artery signi cantly reduced intimal thickening, and this effect was in part mediated by endogenous NO. 252 Likewise, the protective effect of angiotensin converting enzyme (ACE) inhibitors, which also block kinin degradation, may be mediated in part by stimulation of the endogenous production of NO by bradykinin. 253,254 Conversely, treatments with exogenous NO by oral administration of the cysteine-containing NO donor SPM-5185 255 or by chronic inhalation of NO 256 were effective in reducing the size of intimal lesions after injury of the rat carotid artery. Nitroglycerin treatment only decreased the initial medial smooth muscle cell proliferation without affecting the thickness of the neointima after 3 weeks. 257 This may be due to insuf cient NO formation as a result of the well known development of tolerance associated with this class of nitrovasodilators. 101 A single local treatment of the denuded rabbit femoral artery with a protein adduct of NO inhibited platelet deposition and neointimal proliferation in the injured rabbit femoral artery. 258 In vivo eNOS gene transfer in the vessel wall after denudation of the rat carotid artery provided further evidence for the inhibition of smooth muscle cell accumulation by NO. Transfection of the eNOS gene in the media not only restored the calcium-dependent NO production and concomitant relaxations of the denuded artery, it also inhibited neointima formation at day 14 after balloon injury by 70%. 259 This experiment provides direct evidence that NO is an endogenous inhibitor of vascular lesion formation in vivo. Furthermore, these experiments suggest the possibility of eNOS transfection or local delivery of long-lived NO adducts as potential therapeutic approaches to treat neointimal hyperplasia.
The inhibition of intimal thickening by NO is not restricted to models characterized by endothelial denudation, but is also seen when intimal thickening is induced in rabbit arteries by the perivascular placement of a collar. Oral treatment with the NO-donor SPM-5185 reduced the collar-induced intimal thickening, whereas only a tendency towards inhibition was observed by treatment with molsidomine, whose active metabolite is the NO donor SIN-1. 260 It is not clear whether the difference between the two drugs was related to the dose, or different characteristics of the NO donors, i.e. the presence of sulphydryl groups in SPM-5185 or the release of superoxide anion from SIN-1.
Finally, there are indications that NO inhibits neointima formation induced by balloon angioplasty of lesion-free animal arteries. The vessel wall distension by the balloon creates a much more extensive injury of the media, predisposition to thrombus formation and accelerated intimal thickening. Although the endothelial cells regenerate quickly, vascular reactivity studies show that the eNOS pathway remains dysfunctional, 134,261 whereas iNOS is induced in non-endothelial vascular cells. 187 Oral L-arginine supplementation improved the endotheliumdependent vasorelaxation and suppressed the intimal hyperplasia after balloon angioplasty of rabbit iliac arteries. 134 In accordance with the nding in the collar model, treatment with the NO donor SIN-1 did not in uence intimal thickening following porcine carotid angioplasty, although the compound was effective in inhibiting medial smooth muscle cell proliferation. 262 Also in two other models of intimal thickening a clear relationship between inhibition of smooth muscle cell mitosis and neointima formation is lacking. Smooth muscle cell mitosis was in uenced less than intimal thickening after eNOS gene transfer in denuded rat arteries 259 and after NO donor treatment of rabbit collared arteries. 260 This suggests that NO exerts its major effect on smooth muscle cell migration, 228,229 which is a crucial event in intimal thickening. Whether inhibition of migration is of importance to human atherosclerosis remains to be determined, as atherosclerosis develops in an existing intima 1, 3 and migration of smooth muscle cells from media to intima is not considered a major determinant in atherogenesis. 4 (Table 3)

LDL oxidation
Nitric oxide is a nitrogen-free radical and can initiate lipid peroxidation in LDL in the absence 204,263 or presence 201,264 266 of superoxide anion. In the former case, LDL modi cation appeared to be restricted to an increase in lipid hydroperoxide content 201 without further evolution to a high-uptake form recognized by the scavenger receptor. 264 Nitrite-and NO-oxidized LDL demonstrated the biological properties of minimally oxidized LDL. 263 More extensive LDL oxidation occurred if superoxide anion was present e.g. during the decomposition of SIN-1. 264 Superoxide combines with nitric oxide to form the stronger oxidants peroxynitrite and its decomposition product the hydroxyl radical. 267 269

Cytotoxic effects
The concept that peroxynitrite formation occurs in atherosclerosis is strongly supported by the immunohistochemical demonstration of extensive nitration of protein tyrosines in advanced human lesions. 270 The presence of 3nitro-L-tyrosine, quanti ed in the human brain by high-performance liquid chromatography, is also considered indicative of oxidative stress induced by reactive oxygen intermediates and nitric oxide. 271 Excessive NO synthesis and peroxynitrite formation have been implicated in cytotoxic effects in endothelial cells, 268,272 smooth muscle cells 273 and macrophages. 274 Cell damage results from the inhibition of mitochondrial respiration, aconitase activity and DNA synthesis, as well as from iron loss. 20 On the other hand, nitric oxide-induced p53 accumulation safeguarded against DNA damage through p53-mediated suppression of iNOS gene expression, thus reducing the potential for NO-induced DNA damage. 275 The release of basic broblast growth factor from damaged vascular smooth muscle cells may counteract the toxic effects on the endothelium by stimulating endothelial cell proliferation. 273 In view of these ndings, the protective effects of antioxidants in several models of atherosclerosis may in part derive from the prevention of NO breakdown by oxygen radicals.

Induction of apoptosis
Nitric oxide has also been reported to cause apoptosis or programmed cell death in macrophages 276,277 and smooth muscle cells. 278 Apoptosis participates in the regulation of the cellularity of intimal lesions in balloon-injured arteries 279 and human atherosclerosis. 280 Theoretically, augmentation of apoptosis by NO could retard plaque growth, which may be considered bene cial. However, an imbalance between proliferation and apoptosis has been suggested to underlie the development of the cell-poor, necrotic core. 281 The size of the core determines the stability of the plaque. Stimulation of apoptotic cell death by NO or other molecules may thus increase the risk of plaque ssure and thromboembolic complications.

Matrix breakdown
Enhanced matrix breakdown by the activation of matrix metalloproteinases by NO 282,283 or the inactivation of the tissue inhibitor of metalloproteinase-1 by peroxynitrite 284 may contribute to the destabilization of the lesions and may promote the development of a necrotic core in advanced plaques.
In summary, it has been known for a decade that the loss of endothelial NO production impairs endothelium-dependent dilatation and promotes vasospasm in atherosclerotic arteries. More recent evidence indicates that dysfunction of the endothelial NO pathway may promote atherosclerosis in view of the described protective effects of NO against leukocyte adhesion, oxidative processes, smooth muscle cell migration and proliferation. On the other hand, there is ample evidence to consider NO as a molecular aggressor in chronic in ammatory processes like atherosclerosis. 285